Survey-mode measurements and analysis T. A. Herring R.W. King M. A. Floyd Massachusetts Institute of Technology GPS Data Processing and Analysis with GAMIT/GLOBK/TRACK UNAVCO Headquarters, Boulder, Colorado 10–14 August 2015 Material from R. W. King, T. A. Herring, M. A. Floyd (MIT) and S. C. McClusky (now ANU)
Measurement Strategies I: Occupation Time Given time and personnel constraints, what are the trade-offs between between spatial and temporal density? Ideally, you would like for the white-noise position uncertainty for an occupation to contribute to the velocity uncertainty at a level less than the usually dominant long- period correlated noise. Typical white-noise uncertainties (Horizotnal and Vertical) as a function of occupation time: 6-8 hrs: mm H, 5-10 mm V hrs: mm H, 3-5 mm V hrs: mm H, 2-4 mm V Observations over 3 or more days will give you more redundancy Observations of 5 or more days will be necessary for mm-level vertical uncertainties If your region has few continuous stations, you should consider running one or two survey-mode stations for the entire time of the survey to provide continuity
Precision v session length for network processing
Measurement Strategies II: Monuments and Instrumentation Issues in site and antenna selection : Monument stability Accessibility Ease of setup Multipath Vandalism There is no clear prescription for all cases; let’s look at some examples
Three primary mounting options Tripod with optical or physical plummet Spike mount Mast Courtesy UNAVO web page
Low mount in a good enviroment STVP Steven’s Pass, Cascades Range in western Washington 18-cm spike mount No long-term repeatability yet, but 44 hrs of observations in 2012 give formal uncertainties 0.5 mm horizontal, 3 mm vertical. Note minimal long-period signal Scattering.
Low mount in a dirty environment B059 Roadside meadow in western Washington 12.5-cm spike mount Two 24-hr measurements in 2012 agree at 1 mm horizontal, 4 mm vertical though the formal uncertainties are 2 mm, 10mm due to high random noise (diffuse multipath or water vapor?) Note minimal long-period signal scattering. Long-term scatter is 3 mm horizontal, 5 mm vertical (monument instablity?)
High mount in a dirty environment C033 Old survey mark in dirt in central Washington. Tripod mount. (Train blockage was short- lived) hr session and 5-hr session agree at 1.5 mm horizontal, 3 mm vertiical. Long-term repeatability 2 mm horizontal, 12 mm vertical. Surprisingly little short-period multipath (dry dirt?)
Low mount on a slope LYFR Rocky river bank in eastern Oregon 12.5-cm spike mount Single 14-hr session. Long- period multipath due to slope and/or reflective rocks ?
Special Characteristics of Survey-mode Data Editing is critical: every point counts Usually combined with cGPS data to provide continuity and a tie to the ITRF Appropriate relative weighting needed in combining with cGPS data Antenna meta-data may be more complicated Heights may be problematic if different antennas used Seasonal errors behave differently than in cGPS data: best strategy is to observe at the same time of the year (unlike cGPS, which has minimal seasonal sensitivity at 1.5, 2.5, 3.5 ….years total span)
Analysis Strategy Generate time series and aggregated h-files for each survey – Use spans less than ~ 30 days to avoid biasing the position estimate from an in correct velocity – Include cGPS data only on days when sGPS data are available to maintain common-mode cancellation – Strenghens the sGPS positions estimates within each survey and allows better assessment of the long-term statistics – Edit carefully the daily values within each span Generate time series and a velocity using the aggregated h-files from a span of 3 or more years – Edit carefully the long-term time series – Add 0.5 of white noise (sig_neu) to the cGPS estimates from each span to avoid overweighting the cGPS position estimates – Use a separate (ie.g PBO) analysis of the daily cGPS time series to get the appropriate RW (mar_neu) values for each cGPS site; use the median RW for the sGPS sites. See sGPS_recipe.txt for detailed commands
Editing example